Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
Drexel University or The Math Forum.


Luis A. Afonso
Posts:
4,725
From:
LIsbon (Portugal)
Registered:
2/16/05


What really matters is the NET difference
Posted:
Jun 12, 2014 4:02 PM


What really matters is the NET difference
A numerical illustration. ONCE upon a time . . . An industrial orchard was formed by 2 apple species. It was found that out of 1000 trees, species A, 53 were contaminated by a worm, 17 species B. ______ pA= 0.053, pB= 0.017 Pooled proportion: 70/2000= 0.035, Z=(0.0530.017)/sqrt(0.035*0.965*2/1000)= 0.036/ 0.00821888= 4.380 We conclude that there are a significant difference between species because pvalue = 0.0000059 <0.05. Now in order to get Z0=1.645, because prob. (Z0< 1.645) = 0.05, we perform: (0.036d)/sqrt =1.645, d = 0.0361.645*0.00821888= 0.02248 And the conclusion is: It was found that there is a 95% probability that d = 0.022 as the NET proportions difference between species A and B to be attacked.
This example proves that NHST could be pushed beyond the NILNULL that seems to be the greatest Psychologist´s concerns at all . . . They (possibly) were not educated at this direction, but has no excuse, at all, they never thought about as Professionals. Against all expectations they put in action a Taliban´s campaign to ban the NHST´s which is a serious, very serious action, the worse they could even carry on (and the stupidest, I should agree). Luis A. Afonso



